The results

Since January 2026 I have run YouTube for a large exam-prep education brand in India. I own the strategy, the packaging and the reporting for its main channel and its English-language channel.

Same 250 days, 2025 against 2026

YouTube Studio exports, 1 January to 7 September in each year.

The turnaround

I joined in January. February was the channel’s lowest month in a year, and the back catalog was fading on its own. With no extra hands in editing or design, the one lever I had was output.

• Uploads went from 416 to 1,198 with the same editing and design resources.
• Most of the extra volume went into Shorts, whose views grew 5.1x. Livestreams went from 108 to 190.
• Thumbnail testing became policy in January, with three variants on most videos. Long-form videos clearing 5 percent click-through went from 17 to 36.
• In July I took a ten-part growth plan into a management review. From then on the channel ran at three times its February daily rate, 39 percent above the pace we committed to.

I chose reach over per-video averages. I would make the same call again. Had new content only repeated 2025, the channel would have ended about 8 percent down. It ended up 19.

Running 13 faculty on camera

The faculty are teachers first. A script that reads like marketing gets rewritten on camera or skipped, so I write briefs they can teach from.

• In August I sent 124 briefs to 13 faculty plus a guest lane: 61 Shorts, 31 long-form, 18 promo Shorts and 14 livestreams.
• Each brief carries a search title, a hook and the same two calls to action, written straight into that person’s live document.
• I guide the editing and design team day to day.
• In June I handed day-to-day operations to a junior executive, with a written responsibility map and a 14-page handbook.
• Since then I have run three free full-length mock-test events in nine weeks, paired 678 students across 4 rounds, and screened 232 event submissions in one pass.

Systems that run without me

No number goes to management until a second code path has produced it too. I built all of these myself.

• A weekly performance tracker, 17 runs so far, that blocks the report until both paths agree.
• A monthly faculty ranking, 6 reports, with views split fairly when two faculty share a video.
• A daily competitor report that ranks 5 ideas from 42 channels before the team starts. 52 daily and 8 weekly reports so far.
• A 384-video playlist rebuilt end to end: 1,586 chapters written from transcripts, all 384 checked on the live page. About 300 hours of manual work by my estimate; it reruns in about 90 minutes.
• 20 interview episodes turned into 67 short-form cut sheets, routed by language to two channels.
• A zero-cost tracker kit that sets up a new exam vertical in a day. Three other teams have it.

I can’t show yet whether the playlist rebuild moved views. That playlist was already growing, so separating a lift from the trend will take quarters.

Research that sets the calendar

• One study pulled 38,682 competitor videos for one exam vertical and became a 14-week publishing plan.
• Others covered 37 channels, 11 channels with 6,687 videos tagged, and 13 channels, plus a 665-video teardown of one competitor.
• The plan for the next cycle has 120 ideas, each tied to a competitor video that already won and a search term people type.

A second-language channel from zero

• The brand’s English-language channel had gone quiet. I restarted it in January from a 13-channel audit and a 13-week calendar.
• In a year it reached about 3,000 subscribers and 355,000 views, and it now feeds a paid course.
• I can’t say how many enrollments it caused, because the funnel doesn’t record where a student first found us.

Recognition

The founder recognized the work at the June all-hands. In August I was Employee of the Month.

Figures were re-derived from YouTube Studio exports two ways before they went to management. Client name on a call.